This study assessed the validity, intra‐rater and inter‐rater reliability of segmentation of in vivo medial gastrocnemius (MG), lateral gastrocnemius (LG) and soleus (SOL) muscle volume measurement using a single sweep freehand 3D ultrasound (3DUS) in children with cerebral palsy (CP). The MG, LG and SOL of both limbs of 18 children with CP (age 8 years 4 months ± 1 year 10 months, 11 males, unilateral CP = 9, bilateral CP = 9, Gross Motor Functional Classification System I = 11, II = 7) were scanned using freehand 3DUS and magnetic resonance imaging (MRI). All freehand 3DUS and MRI images were segmented and volumes rendered by two raters. Validity was assessed using limits of agreement method. Intra‐rater and inter‐rater reliability was assessed using intra‐class correlation (ICC), coefficient of variance (CV) and minimal detectable change (MDC). Freehand 3DUS overestimated muscle volume of the MG and LG by < 0.3 mL (1%) and underestimated SOL by < 1.3 mL (1.5%) compared with MRI. ICCs for intra‐rater reliability of the segmentation process for the freehand 3DUS system and MRI for muscle volume were > 0.98 and 0.99, respectively, for all muscles. ICCs for inter‐rater reliability of the segmentation process for freehand 3DUS and MRI volumes were > 0.96 and 0.98, respectively, for all muscles. MDCs for single rater freehand 3DUS and MRI were < 4.0 mL (14%) and 3.2 mL (11%), respectively, in all muscles. Freehand 3DUS is a valid and reliable method for the measurement of lower leg muscle volume that can be measured with a single sweep in children with CP in vivo. It can be used as an alternative to MRI for the detection of clinically relevant changes in calf muscle volume as the result of growth and interventions.